Autosomal dominant cerebellar ataxia type I Clinical features and MRI in families with SCA1, SCA2 and SCA3

Brain (1996), 119, 1497-1505

Autosomal dominant cerebellar ataxia type IClinical features and MRI in families with SCA1, SCA2 andSCA3

K. Biirk,1 M. Abele,1 M. Fetter,1 J. Dichgans,1 M. Skalej,2 F. Laccone,3 O. Didierjean,4 A. Brice4 andT. Klockgether1

Departments of ^Neurology and 2Neuroradiology, Correspondence to: Dr T. Klockgether, Department ofUniversity of Tubingen, the ^Department of Human Neurology, University of Tubingen, Hoppe-Seyler-Strafie 3,Genetics, University of Gottingen, Germany and 4INSERM D-72076 Tubingen, GermanyU 289, Hopital de la Salpetriere, Paris, France

SummarySixty-five patients suffering from autosomal dominantcerebellar ataxia-I(ADCA-l) were subjected to a genotypephenotype correlation analysis using molecular geneticassignment to the spinocerebellar ataxia type 1, 2 or 3(SCA1, -2 or -3) locus, clinical examination, eye movementrecording and morphometric analysis of MRIs. Pyramidaltract signs, pale discs and dysphagia were more frequent inSCA1 compared with SCA2 and SCA3 patients. Saccadevelocity was reduced in 56% ofSCAl and all SCA2, but onlyin 30% of SCA3 patients. MRIs of SCA2 patients showedatrophy changes typical of severe olivopontocerebellaratrophy (OPCA). The morphological changes in SCA1 were

similar but less pronounced. In contrast, SCA3 patients hadonly mild cerebellar and brain stem atrophy distinct fromtypical OPCA. The principal finding of this study is thatmutations of the SCA2 and SCA3 gene cause phenotypeswhich can be distinguished in vivo by recording of eyemovements and morphometric MRI analysis. Correlativeplotting of saccade velocity and diameter of the middlecerebellar peduncle yields a clear separation of SCA2 andSCA3. Spinocerebellar ataxia type 1 falls into an intermediaterange that overlaps with both SCA2 and SCA3. However, theclinical syndrome observed in SCA1 patients is different fromthat in SCA2 and SCA3.

Keywords: autosomal dominant cerebellar ataxia; saccade velocity; trinucleotide repeat

Abbreviations: ADCA = autosomal dominant cerebellar ataxia; EOG = electrooculography; MJD = Machado-Josephdisease; OPCA = olivopontocerebellar atrophy; SCA1, -2 or -3 = spinocerebellar ataxia type 1, 2 or 3

IntroductionThe ADCAs are a heterogeneous group of dominantlyinherited disorders characterized by progressive ataxia thatresults from degeneration of the cerebellum and its afferentand efferent connections. Although manifestations ofcerebellar disease are predominant in ADCA, there is oftenclinical and neuropathological evidence for involvementof brainstem, basal ganglia, spinal cord, retina or PNS.Degeneration of one or more of these anatomical structuresmay be present in most families (Greenfield, 1954;Harding, 1982).

Classifications of ADCA have been unsatisfactory as longas the underlying genetic defects were unknown. Tradition-ally, classifications were based on neuropathological criteria.Thus, Holmes (1907) distinguished between spinocerebellar

© Oxford University Press 1996

degeneration, degeneration of the cerebellar cortex and OPCA.More recently, a clinical classification introduced by Hardinggained wide acceptance. Harding separated ADCA into severaltypes, the most common of which, ADCA-I, is characterizedby supranuclear ophthalmoplegia, optic atrophy, basal gangliasymptoms, dementia and amyotrophy. Autosomal dominantcerebellar ataxia-II is distinct in having the additional featureof retinal degeneration, whereas ADCA-III is characterized bya pure cerebellar syndrome (Harding, 1982). Machado-Josephdisease (MJD) is a dominantly inherited ataxic disorder withlarge phenotypic variation. Machado-Joseph disease was firstobserved in patients of Azorean descent (Rosenberg et al.,1976; Coutinho and Andrade, 1978). Although MJD patientsmay have clinical features, such as prominent eyes, severe

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from

1498 K. Biirk et al.

dystonia and amyotrophy, which are less frequently found inNorth American and European ADCA families, there is noclinical evidence to separate MJD from ADCA-I (Harding,1982).

Genetic heterogeneity of ADCA-I has been established,with disease loci assigned to chromosome 6p (SCA1), 12 q(SCA2), 14q (SCA3) and 16q (SCA4) (Zoghbi et al., 1991;Gispert et al., 1993; Gardner et al., 1994; Stevanin et al.,1994). Machado-Joseph disease families have been mappedto a locus on chromosome 14q, coincident with thelocalization of SCA3 (Takiyama et al., 1993). Two of thegenes (SCA1, MJD) have been isolated, and the mutationshave been shown to be unstable trinucleotide (CAG) repeatexpansions present within coding regions of the respectivegenes (Orr et al., 1993; Kawaguchi et al., 1994). The MJDmutation occurs also in European and North American SCA3families, confirming that the genetic basis of SCA3 and MJDis the same (Haberhausen et al., 1995; Matilla et al., 1995).In SCA1 and SCA3, there is an inverse correlation betweenthe length of the CAG repeat and the age of onset, with thelargest alleles occurring in patients with juvenile diseaseonset (Orr et al., 1993; Kawaguchi et al., 1994). In SCA3,patients with marked pyramidal and basal ganglia signs tendto have larger alleles than patients with less severe disease(Maciel et al., 1995).

The identification of various disease loci underlyingADCA-I raises the question whether the different mutationsare associated with distinct clinical phenotypes. Initial studiesfailed to establish clinical differences between SCA1 andSCA3 (Dubourg et al., 1995). Studies describing the clinicalfeatures of SCA2 emphasized the high prevalence of saccadeslowing and the absence of pyramidal tract involvement(Orozco et al., 1990; Lopes-Cendes et al., 1994; Diirr et al.,1995). Direct clinical comparisons of SCA2 with SCA1 orSCA3, however, have not been reported. In this study, wecompared the clinical presentation of SCA1, SCA2 and SCA3families. We used electrooculography (EOG) to measuresaccade velocity and quantitative analysis of MRI to studybrain morphology in vivo.

Patients and methodsPatientsTwenty-one families with a molecular diagnosis of SCA1,SCA2 or SCA3 were selected from the Tubingen ataxiadatabase. At present, this database includes 48 ADCAfamilies. All patients (n = 65) fulfilled diagnostic criteria ofADCA-I which were (i) progressive, otherwise unexplainedataxia in association with at least one of the following signs:saccade slowing, ophthalmoplegia, spasticity, extensor plantarresponses, decreased vibration sense, or dystonia; and (ii)autosomal dominant inheritance. All patients were personallyinterviewed and examined clinically by one of us (K.B.)using a standardized examination procedure. Severity ofcerebellar symptoms was rated on a scale ranging from zero

(absent) to five (most severe) (Klockgether et al., 1990).Electrooculography and MRI were performed in 29 patients.For comparison, two groups of age- and sex-matched healthyvolunteers were studied (EOG, n = 30, age, 47.2± 2.7 years;MRI, n = 36, age, 46.1 ±2.4 years).

Molecular geneticsFor the CAG amplification of the SCA1 region, each 0.025ml reaction contained 60 nM of the SCA1-FP (5'FAM-CAGCTGGAGGCCTATTCCACTCTG-3') and SCA1-1952(5'-TGATGAGCCCCGGAGCCCTGCTGAGGT-3') primers(Orr et al., 1993). For the CAG repeat amplification of theSCA3/MJD gene, each 0.025 ml reaction contained 800 nMof the MJD-2-FP (5'-FAM-TTGATTCGTGAAACAAT-GTATTTT-3') and of the MJD25 (5'-TGGCCTTTCA-CATGGATGTGAA-3') primers (Kawaguchi et al., 1994). Inaddition, each reaction contained 0.08 mM dNTPs (SCA3/MJD: 0.16 mM), 1.5 mM MgCl2, polymerase buffer, 10%dimethylsulphoxide and 200 ng genomic DNA. After a firstdenaturation step at 94°C for 5 min, 1.5 U of TAQ polymerasewas added to the reaction mix. The cycling conditions for35 cycles were, as follows: denaturation at 94°C for 45 s,annealing at 65°C for 30 s (SCA3/MJD: 51°C), elongationat 72°C for 5 min followed by a terminal elongation step at72°C for 5 min. The size of the polymerase chain reactionproducts was determined on an automatic analyser (Genescan,ABI, Foster City).

All subjects from the families without CAG expansion atthe SCA1 or SCA3/MJD loci were genotyped for threemicrosatellite markers which span 2cM on chromosome12 q: cen-D12S105-lcM-D12S1339(1328)-lcM-D12S1340(1329)-tel (Gispert et al., 1995; Krauter et al., 1995).Genotypes were determined by the polymerase chain reaction/blotting technique of Hazan et al. (1992), with slight modi-fications. TAQ polymerase (1.0 U) was added during the firstdenaturation step. Samples then underwent 35 cycles ofdenaturation at 94°C for 15 s, annealing at 55°C for 15 sand elongation at 72°C for 15 s followed by a terminalelongation step of 2 min. Pairwise lod scores were calculatedusing the MLINK program of the computer packageLINKAGE (version 5.1) (Lathrop and Lalouel, 1991).Autosomal dominant transmission and a disease genefrequency of 10̂ * were assumed. Five liability classescalculated according to Ott (1991) were used to take intoaccount the age-dependent penetrance: 0-14 years, 0.075;15-30 years, 0.3; 31^*4 years, 0.574; 45-60 years, 0.803;60 years, 0.955. Allele frequencies were computed from atleast 27 unrelated Caucasian controls. Equal recombinationfractions were assumed for men and women.

ElectrooculographyEye movements were recorded by standard DC EOG methodswith a quasi infinite time constant using silver-silver chlorideelectrodes. Data were written to strip charts. Saccade velocity

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from

Autosomal dominant cerebellar ataxia 1499

(° s ') was calculated by manually determining the maximalslope of the eye position signal of horizontal saccades.Average values of eight 20° centripetal and centrifugalsaccades in both directions made to single light targets,are reported.

MRIMRI was performed using a 1.5 T superconducting system(Magnetom, Siemens AG, Erlangen, Germany) with acircularly polarized head coil. Data were acquired anddisplayed on a 256x256 matrix. A standard examinationprogram was used which consisted of the followingmeasurements: sagittal and axial T,-weighted spin echoimages (TR, 600 ms; TE, 15 ms; NEX 1), axial protondensitiy and T2-weighted spin echo images (TR, 1800 ms,TE, 45/90 ms; NEX 1). All images had a slice thickness of4 mm and were acquired without a gap by two subsequent,interleaved measurements. For quantitative evaluation animage analyser and a computerized interpretation programusing software to overcome partial volume effects was used(Wiillner et al., 1993). For each case, the individual tissuesignal intensity and the CSF signal intensity were defined atthe fourth ventricle and the cerebellar hemispheres,respectively. These measurements gave two sets of normaldistributed data. To determine exactly the amount of tissuewithin an interactively defined region of interest, all pixelswith tissue signal intensity higher than the mean tissue valuewere summed. Those voxels with signal intensities betweenthe mean tissues value and the mean CSF value were weighedaccording to their tissue content to take partial volume effectsinto account. Hereby, an effective number of tissue voxelswithin the region of interest was calculated. This techniquepermits measurement of the amount of tissue within a regionof interest without the necessity to outline a border betweentissue and CSF which is particularly difficult for thecerebellum. To compensate for individual variations of headsize, we measured the total area of the posterior fossa on themidsagittal plane and related each area measurement to therespective posterior fossa area.

The following areas were measured: cerebellar vermis onthe midsagittal plane; at the level of the aqueduct; cerebellarhemispheres on the first parasagittal plane lateral to the middlecerebellar peduncle; fourth ventricle on the midsagittal plane;pontine base with the medial lemniscus as dorsal border onthe midsagittal plane; medulla oblongata at the level of theinferior olivary complex on a horizontal plane; and thecervical spinal cord on a horizontal plane at the level ofthe dens. The maximum diameter of the middle cerebellarpeduncle was measured on a horizontal plane using a distancealgorithm.

All quantitative measures are expressed as a percentage ofthe mean control group value. A morphological diagnosis ofOPCA was made, if the following criteria were fulfilled: (i)size of the cerebellar vermis or hemispheres below the 2 SDrange of the control group; (ii) size of at least two within

the following three anatomical structures below the 2 SDrange of the control group (pontine base, medial cerebellarpeduncle, medulla oblongata) (Wiillner et al., 1993).

Statistical analyisStatistical analysis was performed individual by individual.The relationship between age at onset, saccade velocity,morphometric measures and CAG repeat number wasevaluated through linear regression analysis. Statisticalanalysis of the clinical data was performed using the Kruskal-Wallis test (cerebellar rating) and yj- test (frequency ofassociated symptoms). Statistical differences of saccadevelocity and morphometric measures were calculated byANOVA followed by a Tukey test.

ResultsMolecular genetic analysisAn expansion of the CAG repeat at the SCA1 locus wasdetected in nine affected individuals of eight families. Theexpanded alleles ranged from 47 to 52 repeats, the normalalleles from 28 to 32. Expansion of the CAG repeat at theSCA3 locus was detected in 32 affected individuals of 10families. The expanded alleles ranged from 63 to 80 repeats,the normal alleles from 12 to 36. Linkage analysis wasperformed in three families without expansion at the SCA1or SCA3/MJD loci with three microsatellite markers closelylinked to the SCA2 locus on chromosome 12q. Pairwise lodscores between the disease locus and chromosome 12 markersare shown in Table 1. Positive lod scores were obtained inall families at the loci with a maximal value exceeding +3.0for the combined data. A maximal lod score of +3.25 at arecombination rate of 0.00 for marker D12S105 in family 2established linkage to the SCA2 locus. The families sharedthe same haplotype for the linked markers. Since the expectedfrequency of this haplotype is <1.5%, this result stronglysuggests the existence of a common founder for the threefamilies. A founder effect would not be surprising in thiscase, since the families, although not known to be related,come from the same village in the state of Bavaria. Takentogether, these data indicate that the three families are of theSCA2 type.

Clinical findingsMean age of onset of the examined affected individuals was34.0±4.3 years in SCA1 (n = 9), 37.8±2.6 years in SCA2(« = 24), and 41.5 ±2.3 years in SCA3 (n = 32). Diseaseduration was 10.3±2.5 years in SCA1, 11.4±1.5 years inSCA2, and 10.5 ±1.3 years in SCA3 (Table 2). There wereno significant differences between the groups. Age of onsetwas negatively correlated with repeat length in SCA1 andSCA3 (SCA1: r = -0.76, P < 0.05; SCA3: r = -0.92,P < 0.001).

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from

1500

Table 1

Families

123Total123Total123Total

K. Biirk

Pairwise

0.0

0.263.251.555.061.411.780.443.580.960.580.101.64

et al.

loci scores for

0.01

0.363.181.525.061.391.730.423.540.930.560.101.59

markers on

0.05

0.592.901.374.861.411.530.373.310.810.450.081.34

chromosome 12 q in

0.1

0.712.541.184.431.321.280.302.900.670.340.061.07

three families

0.2

0.691.800.793.281.000.810.171.980.390.170.020.58

0.3

0.511.070.432.010.630.420.071.120.180.050.010.24

0.4

0.270.430.140.840.280.140.020.440.050.000.000.05

Markers

D12S105D12S105D12S105D12S105D12S1339(1328)D12S1339(1328)D12S1339(1328)D12S1339(1328)D12S 1340(1329)D12S 1340(1329)D12S 1340(1329)D12S 1340(1329)

Table 2 Clinical characteristics of SCA1, SCA2 and SCA3patients

Affected examined (n)Age of onset (years)Disease duration (years)Associated signs (%)

Ankle reflexes decreased or absentAnkle reflexes increasedExtensor plantar responseSpasticityAmyotrophyFasciculation-like movementsDecreased vibration senseVertical gaze palsyPale discsProminent eyesDysphagiaBladder dysfunctionDystonia

SCA1

934.0±4.31O.3±2.5

118933781122783356118944

0

SCA2

2437.8±2.611.4±1.5

7 5 "17**2529*3342926310**2154*714

SCA3

3241.5 + 2.310.5± 1.3

63**19**1322**3138695617*2534**4713

Age and duration are given as means SE. *P <0.05; **P < 0.01 versusSCA1.

in SCA2 and SCA3. Parkinsonism was not encountered inany of the patients (Table 2).

Saccade velocitySaccade velocity was moderately reduced in SCA1(244.4±21.8° s-1) and severely reduced in SCA2 (137.9±24.7° s"1). In contrast, it was almost normal in SCA3 (347.719.7° s"1; controls: 383.8±11.8° s"1). Statistical analysisshowed significant differences between SCA2 and all othergroups (P < 0.05 versus SCA1, P < 0.001 versus controlsand SCA3). In addition, SCA1 was different from controls(P < 0.001) and SCA3 (P < 0.01), whereas there was nodifference between SCA3 and controls. On an individualbasis, the saccade velocities of 56% of the SCA1 patients,100% of the SCA2 patients and 30% of the SCA3 patientsfell below the 2 SD range of the control group. There wasno correlation between saccade velocity and disease durationin any of the SCA groups. Similarly, saccade velocity andCAG repeat length were not significantly correlated in SCA1and SCA3.

Mean age of onset of the patients who underwent EOGand MRI (n = 29) was 34.0±4.3 years in SCA1 (n = 9),34.6±4.6 years in SCA2 (« = 8), and 41.3±3.7 years inSCA3 (n = 12). Disease duration was 10.3±1.5 yearsin SCA 1, 8.9± 1.6 years in SCA2, and 7.0± 1.3 years in SCA3.There were no significant differences between the groups.

All patients presented with a cerebellar syndrome. Severityof ataxia of gait, stance, and lower limbs was similar in allgroups. In contrast, upper limb ataxia and action tremor weremore severe in SCA2 compared with SCA3 (P< 0.01), anddysarthria was less severe in SCA3 compared with SCA1and SCA2 (P < 0.01). Increased ankle reflexes (P < 0.01);spasticity (P < 0.05 versus SCA2, P < 0.01 versus SCA3);pale discs (P < 0.01 versus SCA2; P < 0.05 versus SCA3),and dysphagia (P < 0.05 versus SCA2; P < 0.01 versusSCA3) were more frequent in SCA1 than in SCA2 andSCA3. In contrast, decreased or absent ankle reflexes wereless frequent in SCA1 (P < 0.01). Dystonia was a rare finding

MRI morphometryFigure 1 shows representative examples of T|-weighted MRIsof the posterior fossa and cervical spinal cord in SCA1,SCA2 and SCA3. Morphometric MRI analysis revealedsignificant atrophy of the cerebellar vermis, cerebellarhemispheres, pontine base, middle cerebellar peduncle,medulla oblongata, cervical spinal cord and enlargement ofthe fourth ventricle in all SCA groups compared with controls.The only exception was the cerebellar hemispheres in SCA3which were not reduced in size (Fig. 2).

Statistical analysis did not reveal differences betweenSCA1, SCA2 and SCA3 with respect to the size of thecerebellar vermis, the fourth ventricle, the medulla oblongataand the cervical spinal cord. In contrast, the middle cerebellarpeduncles (P < 0.001) and the pontine base (P < 0.01 versusSCA1, P < 0.001 versus SCA3) were significantly smallerin SCA2 compared with SCA1 and SCA3. In addition, the

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from

Autosomal dominant cerebellar ataxia 1501

Fig. 1 T|-weighted MRIs of infratentorial brain structures showing the posterior fossa in the midsagittal plane (upper left) and axialimages at the level of the middle cerebellar peduncles (upper right), inferior olive complex (lower left) and dens axis (lower right).(A) MR1 of 43-year-old healthy male. The areas and distances that were measured are indicated by dashed lines. (B) MRI of a 44-year-old male SCAI patient. (C) MRI of a 38-year-old female SCA2 patient. (D) MRI of a 41-year-old male SCA3 patient.

cerebellar hemispheres were smaller in SCA2 than in SCA3(P < 0.05). There were no differences between SCAI andSCA3 (Fig. 2). Using morphometric criteria (Wiillner et ai.1993), a diagnosis of OPCA was made in 33% of SCAI.75% of SCA2 and in 8% of SCA3 patients. There wasno significant correlation between size of the anatomicalstructures and disease duration in any of the SCA groups.Similarly, size of anatomical structures and CAG repeatlength were not correlated in SCAI and SCA3.

DiscussionIn an attempt to identify characteristic phenotypic featuresof the mutations leading to ADCA-I. we compared patientswith the SCAI. SCA2 and SCA3 mutation. Diagnosis ofSCA I and SCA3 ataxia was made by demonstration of CAGrepeat expansion at the SCAI and SCA3/MJD loci in eightand 10 families, respectively. In three families withoutexpansion at these loci linkage, analysis with markers closelylinked to the SCA2 locus was performed. Although positive

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from

1502 K. Biirk et al.

(A)cerebellar vermis

(B)cerebellar hemispheres

125-

105 -

85 -

65 -

45 -

i

1

•

1

!

+

i

*

jj•

135-

115-

95 -

75-

T-

1

ana

•

—

•i

•

•-f

1

•

i1

•

SCA1 SCA 2 SCA 3 controls

(C) middle cerebellar peduncles120

100 -

SCA1 SCA 2 SCA 3 controls

pontine base

0>N

SCA1 SCA 2 SCA 3 controls SCA1 SCA 2 SCA 3 controls

medulla oblongata ( F ) cervical spinal cord

120-

^ 1 0 0 •

w

N 8 ° '

60-

dn -

i i

•

1

•

•

•

1J.

1ii

•

SCA1 SCA 2 SCA 3 controls SCA1 SCA 2 SCA 3 controls

Fig. 2 Size of cerebellar vermis (A), cerebellar hemispheres (B), middle cerebellar peduncles (C),pontine base (D), medulla oblongata (E) and cervical spinal cord (F) in SCA 1, SCA2, SCA3 andcontrols, as determined by MRI morphometry. The size of each structure is expressed as the percentageof the respective control value. The individual data and the group means are shown. Significances:**P < 0.01, ***P < 0.001 versus controls; +P < 0.05, + + + P < 0.001 versus SCA3; XXP < 0.01,XXXP< 0.001 versus SCA1.

lod scores were obtained in all families, linkage wasestablished only in one family. However, the families camefrom the same village and shared the same haplotype,

suggesting a common founder and a SCA2 genotype inthese families.

The principal and novel finding of this study is that

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from

Autosomal dominant cerebellar ataxia 1503

oo

0

a Kuoa(0

40 50 60 70 80 90

size of middle cerebellar peduncle (%)

Fig. 3 Relationship between saccade velocity and size of themiddle cerebellar peduncles in SCA1 (closed triangles), SCA2(open circles), and SCA3 (plus symbols).

mutations of the SCA2 and SCA3 gene cause phenotypeswhich can be distinguished in vivo by recording of eyemovements and morphometric MRI analysis. Spinocerebellarataxia type 2 is characterized by saccade slowing and MRIchanges typical for OPCA with severe shrinkage of themiddle cerebellar peduncles. In SCA3, saccade velocity isalmost normal and infratentorial atrophy is mild and untypicalfor OPCA. Correlative plotting of saccade velocity anddiameter of the middle cerebellar peduncle yields a clearseparation of SCA2 and SCA3. Spinocerebellar ataxia type1 falls into an intermediate range that overlaps with bothSCA2 and SCA3 (Fig. 3). However, the clinical syndromeobserved in SCA1 patients is different from that in SCA2and SCA3 in that pyramidal tract signs, pale discs anddysphagia are more frequent.

The shared haplotype and the geographical origin of ourSCA2 families suggest that they effectively represent onepedigree. It is therefore of critical importance to questionwhether this pedigree is representative for the SCA2 type ofADCA. Obviously, the present data do not allow a definiteanswer to this question. However, comparison of our familieswith SCA2 families previously reported show a high degreeof resemblance (Orozco et al., 1990; Belal et al., 1994;Lopes-Cendes et al., 1994; Diirr et al., 1995; Filla et al.,1995). A final answer to this question will come from futurestudies including SCA2 families of different geographicaland ethnical origin.

All patients suffered from a pancerebellar syndrome. Ataxiaof upper limbs and action tremor were more severe in SCA2patients than in SCA3 patients indicating major involvementof the cerebellar hemispheres in SCA2. Associated non-cerebellar symptoms may occur with all mutations. Pyramidaltract signs including hyperreflexia and spasticity were morefrequent in SCA1 than in SCA2 and SCA3 patients (Schut,1950; Zoghbi et al., 1988; Goldfarb et al., 1989; Spadaroet al., 1992; Giunti et al., 1994; Dubourg et al., 1995; Geniset al., 1995; Kameya et al., 1995). However, pyramidal tractsigns do not occur exclusively in SCA1. In the present study,there were similar findings in 20-30% of SCA2 and SCA3patients. Earlier studies indicate that pyramidal tract signs

are present in a higher proportion of SCA3 patients, inparticular in those with an early disease onset (Takiyamaet al., 1993; Dubourg et al., 1995; Matilla et al., 1995). Palediscs and dysphagia were frequent signs in SCA1, but didnot occur exclusively in SCA1.

Slow saccades have been observed in several ADCA-Ifamilies (Wadia and Swami, 1971; Orozco et al, 1990).Because it is difficult to assess saccade velocity accuratelyby clinical examination, we used EOG. Although severedisability of some of the patients did not allow extensivestudies of the saccadic system, recording of 20° horizontalsaccades revealed highly significant differences between themutations. Saccade velocity was severely reduced in SCA2while it is usually normal in SCA3 and intermediate inSCA1. There is a clear distinction between SCA2 and SCA3with little overlap of the individual data, whereas saccadevelocity is extremely variable in SCA1 and falls into a rangethat overlaps with both SCA2 and SCA3. Saccade velocityhas previously not been measured in ADCA patients withknown genotype. However, slow saccades have been observedclinically in ~70% of the SCA2 cases of Cuban, Canadian,West Indian and Italian origin (Orozco et al., 1990; Belalet al., 1994; Lopes-Cendes et al., 1994; Diirr et al., 1995;Filla et al., 1995). In contrast, severe saccade slowing hasbeen reported to be absent or is not mentioned in clinicaldescriptions of SCA3 (Takiyama et al., 1994; Dubourg et al.,1995; Maciel et al., 1995; Matilla et al., 1995).

The morphological changes of the cerebellum, brainstemand spinal cord, which underlie the various ataxic disorderscan be studied quantitatively in vivo by MRI morphometry.These studies appear to be of particular importance becauseneuropathological abnormalities have been used for decadesas the only criterion to distinguish between different formsof hereditary ataxia (Holmes, 1907; Greenfield, 1954;Klockgether et al., 1993). The present morphometric MRIdata show an atrophy pattern suggestive of severe OPCA inthe majority of SCA2 patients. As predicted by severe limbataxia, there is often marked atrophy of the cerebellarhemispheres. The infratentorial abnormalities in SCA1 weresimilar but more variable and less pronounced. MRI changessuggestive of OPCA were present in 33% of the SCA1patients. In SCA3, cerebellar and brainstem atrophy was mildand of a pattern incompatible with OPCA. The size of thecervical spinal cord was reduced in all three mutations.Although the number of post-mortem studies in cases withidentified SCA1, SCA2 or SCA3 mutations is limited theresults of these studies agree with the present quantitativeMRI data. Post-mortem examinations of SCA1 cases showedOPCA of variable degree with involvement of ascendingspinal pathways and minor degeneration of the pyramidaltract (Schut, 1950; Greenfield, 1954; Goldfarb et al., 1989;Spadaro et al, 1992; Genis et al, 1995; Kameya et al, 1995).Neuropathological examinations of Cuban SCA2 patientsconsistently revealed OPCA with marked reduction ofPurkinje cells, degeneration of the inferior olives, pontinenuclei, and pontocerebellar fibres. The majority of Cuban

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from

1504 K. Biirk et al.

SCA2 patients had additional degeneration of posteriorcolumns and spinocerebellar pathways, and cell loss in thesubstantia nigra (Orozco et al., 1989). Neuropathologicalfindings in SCA3 are different from those in SCA1 andSCA2. The cerebellar cortex and the inferior olives arespared. The spinocerebellar tracts are most affected, alongwith degeneration of the vestibular and dentate nuclei. Inmost cases, the pontine base is only moderately affected.There is frequent involvement of the substantia nigra andthe subthalamopallidal connections (Rosenberg et al., 1976;Takiyama et al., 1994).

Our data show, not only phenotypical differences betweenthe mutations, but also phenotypic variation within eachmutation. The reasons for this variability are unknown.Correlation analysis failed to establish a relationship betweensaccade slowing or infratentorial atrophy and disease duration.Similarly, correlation of repeat length with saccade velocityand various morphometric variables in SCA1 and SCA3 didnot yield significant results. In contrast, there was a highcorrelation between trinucleotide repeat length and age ofonset, as reported in earlier studies (Orr et al., 1993;Kawaguchi et al., 1994). These data suggest that there aregenetic or non-genetic factors other than trinucleotide repeatlength and disease duration which influence phenotypicvariation within each SCA mutation.

ReferencesBelal S, Cancel G, Stevanin G, Hentati F, Khati C, Ben Hamida C,et al. Clinical and genetic analysis of a Tunisian family withautosomal dominant cerebellar ataxia type 1 linked to the SCA2locus. Neurology 1994; 44: 1423-6.

Coutinho P, Andrade C. Autosomal dominant system degenerationin Portuguese families of the Azores Islands. A new genetic disorderinvolving cerebellar, pyramidal, extrapyramidal and spinal cordmotor functions. Neurology 1978; 28: 703-9.

Dubourg O, Diirr A, Cancel G, Stevanin G, Chneiweiss H, PenetC, et al. Analysis of the SCA1 CAG repeat in a large number offamilies with dominant ataxia: clinical and molecular correlations.Ann Neural 1995; 37: 176-80.

Diirr A, Smadja D, Cancel G, Lezin A, Stevanin G, Mikol J, et al.Autosomal dominant cerebellar ataxia type I in Martinique (FrenchWest Indies). Clinical and neuropathological analysis of 53 patientsfrom three unrelated SCA2 families. Brain 1995; 118: 1573-81.

Filla A, De Michele G, Banfi S, Santoro L, Perretti A, CavalcantiF, et al. Has spinocerebellar ataxia type 2 a distinct phenotype?Genetic and clinical study of an Italian family. Neurology 1995;45: 793^-6.

Gardner K, Alderson K, Galster B, Kaplan C, Leppert M, PtacekL. Autosomal-dominant spinocerebellar ataxia: clinical descriptionof a distinct hereditary ataxia and genetic localization to chromosome16 (SCA4) in a Utah kindred [abstract]. Neurology 1994; 44 Suppl2: A361.

Genis D, Matilla T, Volpini V, Rosell J, Davalos A, Ferrer I,et al. Clinical, neuropathologic, and genetic studies of a large

spinocerebellar ataxia type 1 (SCA1) kindred: (CAG)n expansionand early premonitory signs and symptoms [see comments].Neurology 1995; 45: 24-30. Comment in: Neurology 1995; 45: 1-5.

Gispert S, Twells R, Orozco G, Brice A, Weber J, Heredero L,et al. Chromosomal assignment of the second locus for autosomaldominant cerebellar ataxia (SCA2) to chromosome 12q23—24.1. NatGenet 1993; 4: 295-9.

Gispert S, Lunkes A, Santos N, Orozco G, Ha-Hao D, Ratzlaff-T,et al. Localization of the candidate gene D-amino acid oxidaseoutside the refined 1-cM region of spinocerebellar ataxia 2 [letter].Am J Hum Genet 1995; 57: 972-5.

Giunti P, Sweeney MG, Spadaro M, Jodice C, Novelletto A,Malaspina P, et al. The trinucleotide repeat expansion onchromosome 6p (SCA1) in autosomal dominant cerebellar ataxias.Brain 1994; 117: 645-9.

Goldfarb LG, Chumakov MP, Petrov PA, Fedorova NI, GajdusekDC. Olivopontocerebellar atrophy in a large Iakut kinship in easternSiberia. Neurology 1989; 39: 1527-30.

Greenfield JG. The spino-cerebellar degenerations. Springfield (IL):Charles C. Thomas, 1954.

Haberhausen G, Damian MS, Leweke F, Miiller U. Spinocerebellarataxia, type 3 (SCA3) is genetically identical to Machado-Josephdisease (MJD). J Neurol Sci 1995; 132: 71-5.

Harding AE. The clinical features and classification of the late onsetautosomal dominant cerebellar ataxias. A study of 11 families,including descendants of 'the Drew family of Walworth'. Brain1982; 105: 1-28.

Hazan J, Dubay C, Pankowiak M-P, Becuwe N, Weissenbach J. Agenetic linkage map of human chromosome 20 composed entirelyof microsatellite markers. Genomics 1992; 12: 183-9.

Holmes G. An attempt to classify cerebellar disease, with a noteon Marie's hereditary cerebellar ataxia. Brain 1907; 30: 545-67.

Kameya T, Abe K, Aoki M, Sahara M, Tobita M, Konno H, et al.Analysis of spinocerebellar ataxia type 1 (SCAl)-related CAGtrinucleotide expansion in Japan. Neurology 1995; 45: 1587-94.

Kawaguchi Y, Okamoto T, Taniwaki M, Aizawa M, Inoue M,Katayama S, et al. CAG expansions in a novel gene for Machado-Joseph disease at chromosome 14q32.1 [see comments]. Nat Genet1994; 8: 221-8. Comment in: Nat Genet 1994; 8: 213-15.

Klockgether T, Schroth G, Diener HC, Dichgans J. Idiopathiccerebellar ataxia of late onset: natural history and MRI morphology.J Neurol Neurosurg Psychiatry 1990; 53: 297-305.

Klockgether T, Wullner U, Dichgans J, Grodd W, Nagele T, PetersenD, et al. Clinical and imaging correlations in inherited ataxias.[Review]. Adv Neurol 1993; 61: 77-96.

Krauter K, Montgomery K, Yoon S-J, LeBlanc-Straceski J, RenaultB, Marondel I, et al. A second-generation YAC contig map ofhuman chromosome 12. Nature 1995; 377 Suppl: 321-33.

Lathrop GM, Lalouel JM. Statistical methods for linkage analysis.In: Rao CR, Chakraborty R, editors. Handbook of statistics, Vol. 8.Amsterdam: Elsevier, 1991: 81-123.

Lopes-Cendes I, Andermann E, Attig E, Cendes F, Bosch S, WagnerM, et al. Confirmation of the SCA-2 locus as an alternative locus

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from

Autosomal dominant cerebellar ataxia 1505

for dominantly inherited spinocerebellar ataxias and refinement ofthe candidate region. Am J Hum Genet 1994; 54: 774-81.

Maciel P, Gaspar C, DeStefano AL, Silveira I, Coutinho P, RadvanyJ, et al. Correlation between CAG repeat length and clinical featuresin Machado-Joseph disease. Am J Hum Genet 1995; 57: 54-61.

Matilla T, McCall A, Subramony SH, Zoghbi HY. Molecular andclinical correlations in spinocerebellar ataxia type 3 and Machado-Joseph disease. Ann Neural 1995; 38: 68-72.

Orozco G, Estrada R, Perry TL, Arana J, Fernandez R, Gonzalez-Quevedo A, et al. Dominantly inherited olivopontocerebellar atrophyfrom eastern Cuba. Clinical, neuropathological, and biochemicalfindings. J Neurol Sci 1989; 93: 37-50.

Orozco Diaz G, Nodarse Fleites A, Cordoves Sagaz R, AuburgerG. Autosomal dominant cerebellar ataxia: clinical analysis of263 patients from a homogeneous population in Holguin, Cuba.Neurology 1990; 40: 1369-75.

Orr HT, Chung MY, Banfi S, Kwiatkowski TJ Jr, Servadio A,Beaudet AL, et al. Expansion of an unstable trinucleotide CAGrepeat in spinocerebellar ataxia type 1. Nat Genet 1993; 4: 221-6.

Ott J. Analysis of human genetic linkage. Rev. ed. Baltimore: JohnsHopkins University Press, 1991.

Rosenberg RN, Nyhan WL, Bay C, Shore P. Autosomal dominantstriatonigral degeneration. A clinical, pathologic, and biochemicalstudy of a new genetic disorder. Neurology 1976; 26: 703-14.

Schut JW. Hereditary ataxia. Clinical study through six generations.Arch Neurol Psychiatry 1950; 63: 535-68.

Spadaro M, Giunti P, Lulli P, Frontali M, Jodice C, Cappellacci S,et al. HLA-linked spinocerebellar ataxia: a clinical and geneticstudy of large Italian kindreds. Acta Neurol Scand 1992; 85: 257-65.

Stevanin G, Le Guern E, Ravise N, Chneiweiss H, Diirr A, CancelG, et al. A third locus for autosomal dominant cerebellar ataxiatype I maps to chromosome 14q24.3-qter: evidence for the existenceof a fourth locus. Am J Hum Genet 1994; 54: 11-20.

Takiyama Y, Nishizawa M, Tanaka H, Kawashima S, Sakamoto H,Karube Y, et al. The gene for Machado-Joseph disease maps tohuman chromosome 14q. Nat Genet 1993; 4: 300-4.

Takiyama Y, Oyanagi S, Kawashima S, Sakamoto H, Saito K,Yoshida M, et al. A clinical and pathologic study of a large Japanesefamily with Machado-Joseph disease tightly linked to the DNAmarkers on chromosome 14q. Neurology 1994; 44: 1302-8.

Wadia NH, Swami RK. A new form of heredo-familialspinocerebellar degeneration with slow eye movements (ninefamilies). Brain 1971; 94: 359-74.

Wullner U, Klockgether T, Petersen D, Naegele T, Dichgans J.Magnetic resonance imaging in hereditary and idiopathic ataxia [seecomments]. Neurology 1993; 43: 318-25. Comment in: Neurology1993; 43: 2425-6.

Zoghbi HY, Pollack MS, Lyons LA, Ferrell RE, Daiger SP, BeaudetAL. Spinocerebellar ataxia: variable age of onset and linkage tohuman leukocyte antigen in a large kindred. Ann Neurol 1988; 23:580-4.

Zoghbi HY, Jodice C, Sandkuijl LA, Kwiatkowski TJ Jr, McCall AE,Huntoon SA, et al. The gene for autosomal dominant spinocerebellarataxia (SCA1) maps telomeric to the HLA complex and is closelylinked to the D6S89 locus in three large kindreds. Am J Hum Genet1991; 49: 23-30.

Received March 19, 1996. Revised May 16, 1996.Accepted June 13, 1996

by guest on July 12, 2011brain.oxfordjournals.org

Dow

nloaded from